{"id":"916a53f5-576d-4ea9-ace0-354e668c4ce1","arxiv_id":"1908.07104","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The Koolen-Riebeek, Soicher, and affine-geometry incidence graphs on 486 vertices are all realized by the same rank-9 action of 3^5:(2×M10), unified via the ternary Golay code.","lead":"This paper shows that three known distance-regular graphs on 486 vertices, the Koolen-Riebeek graph, the Soicher graph, and an incidence graph from affine geometry, all come from the same group action of 3^5:(2×M10), with the ternary Golay code explaining the link. A smart generalist should read it because it uncovers hidden symmetries shared by apparently different graphs, a step toward better classification of distance-regular graphs.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The identification of the orbital graph (*) as the Soicher graph depends entirely on an unpublished uniqueness theorem cited as [7]; if that theorem is wrong or inapplicable, the paper's central unification claim does not follow.","rationale":"The reader's weakest assumption is exactly the point I would defend as the most load-bearing: the external, unpublished uniqueness theorem is the only bridge from 'a distance-regular graph with these parameters in this H-action' to 'the Soicher graph'. The rest of the paper has independent support: explicit generators are given, the identification of Σ is checked directly in GRAPE against a design construction, and the Golay-code explanation is a coherent theoretical framework. I do not see an internal contradiction. The absence of shipped scripts makes verification harder but is not itself a mathematical flaw. A direct isomorphism test, or an independent check of [7], would settle the concern. Since this is a verification and reproducibility issue rather than a demonstrated error, the conditional verdict remains appropriate and no change is needed.","tokens_in":9747,"tokens_out":7619,"duration_ms":80341,"concrete_test":"Run an explicit isomorphism check rather than relying only on [7]: with the generators in the Appendix, build the 56-regular orbital graph obtained in Section 3 (or equivalently the graph defined in Section 4 by joining G to the 20 weight-1 cosets and the 36 Type II flats, then taking H-translates), and use GRAPE/nauty's IsIsomorphicGraph to compare it with a graph constructed directly from Soicher's 1993 definition. If the two are isomorphic, the uniqueness assumption is unnecessary for the claim; if they are not, Section 3's identification fails. As a cheaper check, inspect [7] for a complete proof of uniqueness for the intersection array {56,45,16,1;1,8,45,56}.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is in Section 3. After listing the intersection arrays obtained from the rank-9 action of H, the paper states that the array {56,45,16,1;1,8,45,56} 'is precisely that of the second Soicher graph Υ' and concludes: 'Since Υ is the unique distance-regular graph with this intersection array, then that must be the graph we have obtained here.' Every subsequent use of the name 'Soicher graph' for this orbital graph — including the abstract's claim that the same rank-9 action preserves Υ — inherits this step. The uniqueness theorem is attributed only to the unpublished corrections file [7]. If that theorem is wrong, or if its hypotheses do not apply, the rank-9 action may still produce a distance-regular graph with the Soicher parameters but not Soicher's graph. The later construction in Section 4 of a valency-56 graph on the same 486 vertices does not remove this dependence: its identification as Υ again rests on the same uniqueness statement. This is a single point of failure, but it is genuinely load-bearing for the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies three known distance-regular graphs on 486 vertices: the Koolen--Riebeek graph Δ, the second Soicher graph Υ, and the incidence graph Σ of a symmetric transversal design from AG(5,3). The authors show that all three graphs can be obtained as unions of orbitals in the same rank-9 action of the group H = 3^5:(2×M10), and they explain the connection using the ternary Golay code. Explicit permutation generators for H are given in the appendix, and orbit diagrams and distance distribution diagrams are provided. The identification of the orbital graph with the Soicher graph Υ, however, is made only by matching the intersection array and invoking an unpublished uniqueness theorem of Brouwer.","tokens_in":9952,"tokens_out":18420,"duration_ms":176430,"significance":"If the constructions are correct, the paper provides a new unified perspective on three known distance-regular graphs, together with a coding-theoretic explanation and a new induced subgraph observation (the coset graph of the shortened ternary Golay code appears inside Υ). The explicit generators and orbit diagrams are valuable computational data. The main weakness is that the central identification with Υ depends entirely on an unpublished uniqueness result, and several key computational verifications are not reproducible from the manuscript. These issues are fixable and do not invalidate the overall approach, but they must be addressed before the main claim can be considered established.","major_comments":[{"comment":"The orbital graph (*) with intersection array {56,45,16,1;1,8,45,56} is identified with the Soicher graph Υ solely by stating that Υ is the unique distance-regular graph with this array, citing the unpublished corrections file [7]. The theorem is neither stated nor proved, and no direct isomorphism check with Soicher's original construction is provided. Since the abstract and all later references to Υ inherit this step, this is a load-bearing point. Please add a precise statement of the uniqueness theorem with a citable source, or supply a computational isomorphism certificate (including code) between the orbital graph and the graph constructed by Soicher in [18].","section":"Section 3 (table and following paragraph)"},{"comment":"Several computational claims are asserted without reproducible support: the Magma weight-distribution classification of the 81 subspaces into Type I and Type II, the Magma computation of setwise stabilizers, the GAP verification that the Type I incidence graph is the Koolen--Riebeek graph Δ, and the GAP check that the orbital graph is isomorphic to Σ. These are not merely implementation details; they are part of the proof of the paper's central claim. Please provide the GAP and Magma scripts (or detailed pseudocode) together with the resulting output, for example as ancillary files or an appendix.","section":"Section 4 (weight distributions and isomorphisms)"},{"comment":"The sentence \"But this is precisely the construction of the Soicher graph Υ given above\" is not justified explicitly. The graph (*) from Section 3 is obtained as an EdgeOrbitsGraph orbital union, but the reader is not told which orbitals are used. Please state which suborbits of the rank-9 action form the edge set of (*) (presumably those of lengths 20 and 36) and show that the graph defined by making G adjacent to the 36 Type II subspaces and the 20 weight-1 cosets is exactly the union of those orbitals. This will make the construction in Section 4 match the orbital graph (*) in a verifiable way.","section":"Section 4 (paragraph on the valency-56 graph)"}],"minor_comments":[{"comment":"The graph listed as \"K3162\" is not standard notation. If it denotes the complete multipartite graph with 162 parts of size 3, please use a standard notation such as $K_{3[162]}$ and verify that the intersection array {483,2;1,483} corresponds to that graph.","section":"Section 3 (table)"},{"comment":"There is a typo: \"Two important class\" should read \"Two important classes\".","section":"Section 1 (introduction)"},{"comment":"Reference [7] is an online corrections file with no version date or access date. Please provide a full citation, including the URL, a version identifier, and the date the file was accessed, so that readers can locate the exact statement of the uniqueness theorem.","section":"References ([7])"},{"comment":"The weight distributions are given in a compact notation that is ambiguous in the extracted text (e.g., \"01 110 270 ...\"). Consider presenting them as explicit weight enumerators (polynomials or tables), which would make the two classes easier to check.","section":"Section 4 (weight distributions)"}],"recommendation":"major_revision","confidential_remarks":"The paper's central claim is conditional on the unpublished uniqueness theorem for the Soicher parameters. Even if the authors trust Brouwer's online corrections, a formal referee report should not accept a load-bearing step that depends on an uncited, unstated theorem. A direct isomorphism check with Soicher's graph, or a proof of uniqueness, would resolve this. The lack of computational scripts is also a reproducibility concern for a paper whose main results are computational. If these are addressed, the paper would be a useful contribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does something real: it shows that the Koolen–Riebeek graph, the Soicher graph, and the affine-geometry incidence graph all appear as orbital graphs of the same rank-9 action of 3^5:(2×M10), and it explains the connection through the ternary Golay code. That unification is not in the earlier literature. The explicit generators in the appendix and the orbit diagrams make the construction concrete, and the induced-subgraph observation about Λ inside Υ is new and plausible. I believe the central claim is probably correct, but the paper has one load-bearing soft spot and one reproducibility gap.\n\nThe soft spot is Section 3: the orbital graph with array {56,45,16,1;1,8,45,56} is identified as the Soicher graph solely because Brouwer proved uniqueness of that array, and that proof lives only in an unpublished corrections file [7]. If that uniqueness statement is wrong or does not apply, then the orbital graph has the Soicher parameters but need not be Soicher's graph. The later Golay-code construction in Section 4 inherits the same dependence when it calls the resulting valency-56 graph Υ. This is a single point of failure, but it is genuinely load-bearing for the abstract's claim that the same action preserves Υ. The identifications of Δ and Σ are on firmer ground: for Σ there is a direct isomorphism check, and for Δ the orbital construction plus automorphism-group verification is convincing.\n\nThe second issue is reproducibility. Several key steps are asserted as GAP or Magma computations without shipped scripts or logs. For a computational paper, that is a real deficiency. The authors give generators and describe the calculations, which is better than nothing, but a referee cannot re-run the identification without re-implementing everything.\n\nWhat is not a problem: there is no circularity. The constructions start from the group and the Golay code, then compare to known intersection arrays and uniqueness results. No free parameters are fitted, and no target graphs are fed into the construction. Also, citing Brouwer's unpublished file is not itself a crime when the result is widely used, but for this particular uniqueness theorem the authors should either upgrade to a published reference or provide a direct isomorphism check against Soicher's original construction.\n\nWho is this for? Researchers in distance-regular graphs, permutation groups, and Golay-code constructions. It deserves a serious referee, not a desk rejection. My recommendation: send to a good combinatorics journal, and require the authors to make the computations reproducible and to shore up the Soicher identification. If those two things happen, this is a solid paper. If not, the central unification remains conditional.","headline":"A genuinely new unification of three 486-vertex distance-regular graphs under one rank-9 group action, with the Soicher-graph identification leaning on an unpublished uniqueness theorem and the computational checks not shipped.","tokens_in":10465,"tokens_out":2301,"would_cite":true,"duration_ms":25497,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05E30","05C25","20B25","94B25","51E05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper shows that three distance-regular graphs on 486 vertices — the Koolen–Riebeek graph, the second Soicher graph, and the incidence graph of a symmetric transversal design from $\\mathrm{AG}(5,3)$ — are all preserved by the same…","keywords":["distance-regular graph","Koolen–Riebeek graph","Soicher graph","ternary Golay code","symmetric transversal design","affine geometry AG(5,3)","orbital graphs","rank-9 group action"],"falsifier":"Build the rank-9 orbital graph from the appendix generators and compare it directly, via an explicit isomorphism or non-isomorphism computation, with a graph constructed from the original 1993 Suzuki-group construction; if they are non-isomorphic, the claimed identification collapses. Equally decisive would be finding two non-isomorphic distance-regular graphs with intersection array $\\{56,45,16,1;1,8,45,56\\}$, which would disprove the uniqueness statement on which the identification rests.","tokens_in":2103,"feed_emoji":"📐","tokens_out":12277,"duration_ms":187222,"temperature":0.7,"pith_summary":"Three well-studied distance-regular graphs on 486 vertices — the Koolen–Riebeek graph, the second Soicher graph, and the incidence graph of a symmetric transversal design from $\\mathrm{AG}(5,3)$ — are shown to be different orbital graphs of a single rank-9 permutation action of the group $3^5:(2\\times M_{10})$. The paper gives an explicit explanation through the ternary Golay code: the vertices are the code's cosets together with 81 four-flats of the affine geometry, and the split of those flats into 45 Type I and 36 Type II subspaces selects the edge rules for the three graphs. If the claim is right, objects previously reached by separate constructions share one symmetry group, and the Soicher graph gains an explicit, code-theoretic description.","feed_headline":"Three 486-vertex graphs are one rank-9 group action","feed_subtitle":"The Koolen–Riebeek, Soicher, and AG(5,3) incidence graphs all arise as orbitals of the same group, linked by the ternary Golay code.","key_machinery":"The load-bearing object is the rank-9 action of $H \\cong 3^5:(2\\times M_{10})$ on 486 points, whose nine suborbits have lengths $1,2,20,36,40,45,72,90,180$; the edge sets of the three graphs are unions of the corresponding orbitals. The explanation is carried by the ternary Golay code $G \\subset \\mathbb{F}_3^{11}$: the 243 cosets of $G$ form one part of each bipartition, and the 81 ten-dimensional subspaces containing $G$ but not $G+e_0$ form the other. Weight-distribution calculations split those 81 subspaces into 45 Type I and 36 Type II spaces, and the differing incidence rules on the same set of 486 vertices reproduce the intersection arrays $\\{45,44,36,5;1,9,40,45\\}$, $\\{56,45,16,1;1,8,45,56\\}$, and $\\{81,80,54,1;1,27,80,81\\}$.","core_discovery":"On the paper's own terms, the central discovery is that each of the imprimitive distance-regular graphs $\\Delta$, $\\Upsilon$ and $\\Sigma$ is preserved by the same rank-9 action of $H \\cong 3^5:(2\\times M_{10})$ on 486 points, with suborbits of lengths $1,2,20,36,40,45,72,90,180$. Starting from the ternary Golay code $G$ in $\\mathbb{F}_3^{11}$, the 243 cosets of $G$ and the 81 ten-dimensional subspaces containing $G$ but not $G+e_0$ form the vertices; the 45 Type I subspaces give the Koolen–Riebeek graph $\\Delta$ as an incidence graph, the 36 Type II subspaces together with the 20 weight-one cosets $\\pm e_i$ give the Soicher graph $\\Upsilon$, and the full collection of 81 subspaces gives the AG(5,3) incidence graph $\\Sigma$. The identification is confirmed by checking weight distributions, setwise stabilizers, and the resulting intersection arrays.","pith_inferences":["Inference: the three graphs are three edge-rules on one 486-point geometry, so other unions of the nine orbitals could be tested for distance-regularity or association-scheme structure, potentially yielding further graphs with the same automorphism group.","Inference: the 45/36 split of the ten-dimensional subspaces is detected purely by weight distribution; if an analogous split exists for other perfect codes, similar coincidences might occur for other rank-k actions.","Inference: because $\\Upsilon$ contains the shortened-Golay coset graph as an induced subgraph, one could ask whether $\\Upsilon$ is a covering or blow-up of that 243-vertex graph with the Type II flats attached, connecting the construction to existing covering theory.","Inference: a direct computer isomorphism test against the original 1993 construction would settle the identification of the orbital graph with $\\Upsilon$ without relying on the unpublished uniqueness result."],"forward_implications":["The same subgroup $3^5:(2\\times M_{10})$ acts as automorphisms of all three graphs, so any property of the vertex set invariant under that subgroup is simultaneously a property of all three graphs.","The Soicher graph $\\Upsilon$ is realized explicitly from the ternary Golay code: put the 20 weight-one cosets and the 36 Type II subspaces adjacent to the zero coset, then translate by the group action.","The Koolen–Riebeek graph $\\Delta$ is the incidence graph of the cosets against the 45 Type I subspaces, giving a direct code-theoretic construction distinct from the original 45-coclique description.","The induced subgraph of $\\Upsilon$ on the 243 cosets is the distance-transitive graph with intersection array $\\{20,18,4,1;1,2,18,20\\}$, previously known as the coset graph of the shortened ternary Golay code; the observation that it embeds as an induced subgraph of $\\Upsilon$ is new."],"supporting_citations":[{"why":"Supplies the construction and intersection array of the Koolen–Riebeek graph and its description via cosets and 45-cocliques.","marker":"[9]"},{"why":"Defines the Berlekamp–van Lint–Seidel graph as the coset graph of the ternary Golay code, giving the 243-vertex strongly regular graph whose complement gives the halved graphs of Δ.","marker":"[3]"},{"why":"Introduces the second Soicher graph, its intersection array, and its construction from a simple group, which is the reference object for the identification.","marker":"[18]"},{"why":"Contains the unpublished uniqueness proof for intersection array {56,45,16,1;1,8,45,56}, on which the identification of the orbital graph with the Soicher graph depends.","marker":"[7]"},{"why":"Provides the standard theory of imprimitive distance-regular graphs, halving and folding, and lists the induced graph known as (A17) in Section 11.3H.","marker":"[6]"},{"why":"Provides the collapsed-adjacency-matrix technique used to compute intersection arrays of orbital graphs for a vertex-transitive group.","marker":"[17]"},{"why":"Describes the construction of symmetric transversal designs from affine geometry by deleting parallel classes of flats, which underlies the graph Σ.","marker":"[4]"},{"why":"Gives an alternative construction of these incidence graphs from projective space and confirms their distance-transitivity.","marker":"[13]"},{"why":"Documents the uniqueness and perfect-code properties of the ternary Golay code used throughout the construction.","marker":"[15]"}],"fun_headline_variants":["One rank-9 group action unites three 486-vertex graphs","Ternary Golay code explains shared structure of three graphs","Three distance-regular graphs share one group action","Same rank-9 action preserves three 486-vertex graphs"],"cache_read_input_tokens":12672,"weakest_assumption_plain":"The load-bearing assumption is that the graph obtained from the rank-9 action with intersection array $\\{56,45,16,1;1,8,45,56\\}$ really is the second Soicher graph; the paper relies on a uniqueness theorem cited only to an unpublished corrections file, so if that uniqueness theorem fails, the constructed graph might be a different distance-regular graph with the same parameters.","fun_headline_variants_meta":{"raw":{"variants":["One rank-9 group action unites three 486-vertex graphs","Ternary Golay code explains shared structure of three graphs","Three distance-regular graphs share one group action","Same rank-9 action preserves three 486-vertex graphs"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00083,"raw_usage":{"total_tokens":3598,"prompt_tokens":893,"completion_tokens":2705,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":509,"completion_tokens_details":{"reasoning_tokens":2637}},"tokens_in":509,"tokens_out":2705,"duration_ms":18677,"temperature":1.0,"reasoning_tokens":2637,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:26:02.800468+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build the rank-9 orbital graph from the appendix generators and compare it directly, via an explicit isomorphism or non-isomorphism computation, with a graph constructed from the original 1993 Suzuki-group construction; if they are non-isomorphic, the claimed identification collapses. Equally decisive would be finding two non-isomorphic distance-regular graphs with intersection array $\\{56,45,16,1;1,8,45,56\\}$, which would disprove the uniqueness statement on which the identification rests.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the construction and intersection array of the Koolen–Riebeek graph and its description via cosets and 45-cocliques."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the Berlekamp–van Lint–Seidel graph as the coset graph of the ternary Golay code, giving the 243-vertex strongly regular graph whose complement gives the halved graphs of Δ."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the second Soicher graph, its intersection array, and its construction from a simple group, which is the reference object for the identification."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Contains the unpublished uniqueness proof for intersection array {56,45,16,1;1,8,45,56}, on which the identification of the orbital graph with the Soicher graph depends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the standard theory of imprimitive distance-regular graphs, halving and folding, and lists the induced graph known as (A17) in Section 11.3H."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the collapsed-adjacency-matrix technique used to compute intersection arrays of orbital graphs for a vertex-transitive group."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Describes the construction of symmetric transversal designs from affine geometry by deleting parallel classes of flats, which underlies the graph Σ."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives an alternative construction of these incidence graphs from projective space and confirms their distance-transitivity."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents the uniqueness and perfect-code properties of the ternary Golay code used throughout the construction."}],"review_version":1}